Blueprinting, reliability, validity and post-test item analysis—the four foundations of a fair and defensible examination.
Four Core Concepts
Concept 01
Table of Specifications (TOS) / Test Blueprint
Core principle/rule: A Table of Specifications—also called a test blueprint—is a two-way chart aligning course content with instructional objectives or cognitive levels so the test is balanced and representative.
Key facts
Axes of the chart: Rows list the specific topics or content areas covered in class; columns list cognitive levels being tested—for example, Bloom's Knowledge, Comprehension and Application.
Primary purpose: Guarantees Content Validity by ensuring comprehensive syllabus coverage in correct proportions.
Table of Specifications — Example
Topics / Content
Cognitive Levels (Bloom's)
Knowledge
Comprehension
Application
1. Algebra
2
2
1
2. Geometry
1
2
2
3. Trigonometry
1
1
2
Total Items
4
5
5
Numbers indicate the number of items planned in the test.
Concept 02
Reliability and Its Estimation Methods
Core principle/rule: Reliability is the consistency and stability of test scores. A reliable test should give roughly the same score when repeated under the same conditions.
Key facts
Test–Retest Method: Give the same test to the same group after a time interval; measures the coefficient of stability.
Split-Half Method: Give the test once, divide it into equal halves—such as odd versus even questions—and correlate their scores; measures internal consistency.
Spearman–Brown Prophecy Formula: Because splitting a test reduces reliability, this estimates whole-test reliability from the correlation between the two halves.
Spearman–Brown Prophecy Formula
rtt = 2r12 / (1 + r12)
r12 = correlation between the two halves; rtt = reliability of the whole test
Concept 03
Types of Validity
Core principle/rule: Validity is accuracy—whether a test actually measures what it claims or intends to measure.
Key facts
Content Validity: Does the test cover a representative sample of the syllabus? Secured through a Table of Specifications.
Criterion-Referenced Validity: How well do scores correlate with an external standard?
Concurrent validity: Measured against a standard at the same time—for example, a new short mathematics test and a long established test taken on the same day.
Predictive validity: Measured against a future standard—for example, whether engineering-entrance scores predict high college GPA.
Construct Validity: Does the test measure an abstract psychological or theoretical construct such as intelligence, anxiety or creativity?
Concept 04
Item Analysis — Difficulty and Discrimination Indices
Core principle/rule: Item analysis is a statistical technique used after test administration to evaluate individual questions and decide whether each should be kept, revised or discarded.
Difficulty Index (P-Index)Proportion of students answering correctly. Range: 0.0 to 1.0. Ideal: usually 0.40–0.60, meaning 40%–60% correct.
Discrimination Index (D-Index)How well an item distinguishes high- from low-scoring students. Range: −1.0 to +1.0. Ideal: high positive, about +0.40 or higher; top students answer correctly and bottom students incorrectly.
Quick Reference
Psychometric Metrics Summary
Swipe sideways to view all columns →
Metric / Concept
What It Measures
Target / Ideal Value
Key Warning Sign
Reliability
Consistency of scores over time or halves.
As close to +1.0 as possible.
A test can be perfectly reliable without being valid.
Validity
Accuracy—does it measure the right thing?
High correlation with the intended criterion.
A test lacking validity is useless, regardless of reliability.
Item Difficulty (P)
Proportion answering correctly.
Around 0.50 for a standard norm-referenced test.
P = 0.95 is too easy; P = 0.10 is too hard or guessing-prone.
Item Discrimination (D)
Ability to separate high versus low overall scorers.
+0.40 or above—excellent.
Negative D-values mean the question is fundamentally flawed.
Test Construction & Standardization · Psychometric quick revision